Entity Extraction is a Critical Technology for Detecting and Monitoring Geopolitical Events

Organizations Need to Know What’s Going on in Their Areas of Operation

In our deeply interconnected world, geopolitical events have ripple effects far beyond their immediate location, affecting both private and public sectors. Think imperiled overseas personnel and facilities, disrupted supply chains, rising fuel prices, large inflows of migrants fleeing violence, terrorist radicalization, etc.

An oil company or an international shipping company or many other kinds of organizations need to know what’s going on in places of vital importance to them, which could very negatively affect their businesses. Governments need to protect their citizens overseas and their policy and security interests.

The large and increasing number of dangerous geopolitical events at present, such as Hamas’ attack on Israel, are only underscoring the need to stay on top of them. Many companies and organizations monitor news, social media, and other data sources to identify geopolitical events that could affect their businesses, assets, and operations.

The vast majority of relevant information is contained in unstructured text data such as news articles or social media posts, but the volumes of data to sift through are enormous, making it difficult if not impossible for a purely human team to handle it all, let alone in real time. It requires an automated assist.

Entity Extraction for Monitoring Geopolitical Events

Fortunately, there’s an AI technology called Entity Extraction that can help. Entity Extraction finds the geopolitical information in very large amounts of unstructured text data and turns it into structured data that other automated processes such as visualization and mapping tools can make use of for actionable insights.

As a first step, Entity Extraction finds the names of entities, such as people, organizations, facilities, time expressions, locations, and monetary amounts. In a keyword search tool like Google, to find a person name you have to know that name and enter it into the tool. Entity Extraction finds all instances of names, whether previously known or unknown.

Entity Extraction does this by performing a sophisticated NLP (natural language processing) analysis of the text surrounding a name and using distinctive clues that indicate the presence of a name. For more information on Entity Extraction, see here.

Going beyond Names, Entity Extraction Also Finds Events and Relationships

An advanced capability of Entity Extraction also finds events expressed in unstructured text and converts them into structured data. The events covered include geopolitical ones such as conflicts, political changes, terrorist attacks, and many others.

Entity Extraction identifies the nature of the event, the participants, and the time and location of the event if given (“Who did what to whom, where and when.”). For more information on extracting events, see here.

For instance, given a sentence like “Hamas militants launch a surprise attack on Israeli territory gunning down hundreds of citizens and taking women and children hostage,” Entity Extraction identifies 3 events:

Event Type: ATTACK

Perpetrator: Hamas militants

Target: Israeli territory

Place: Israel

 

Event Type: KILL

Perpetrator: Hamas militants

Target: hundreds of citizens

 

Event Type: HOLD HOSTAGE

Perpetrator: Hamas militants

Target: women and children

 

Entity Extraction also finds links or relationships between entities such as where people or organizations are located and how those people and organizations are related. For more information on relationship extraction, see here.

For instance, in the sentence “Hamas, which was originally a faction of the Muslim Brotherhood, has a long-standing relationship with the Lebanese group Hezbollah”, Entity Extraction recognizes 3 relationships:

Relationship Type: ORG PARENT

Organization: Hamas

Parent Org: Muslim Brotherhood

 

Relationship Type: ORG AFFILIATION

Organization: Hamas

Organization: Hezbollah

 

Relationship Type: ORG NATIONALITY

Organization: Hezbollah

Country: Lebanon

 

An Advanced Feature of Entity Extraction is its Ability to Geotag an Event

Geotagging enables an extracted event to be plotted on a map. Unstructured text data usually doesn’t contain precise coordinates for locations, but only place names. Some place names are ambiguous (think “Cairo” – is it the capital of Egypt or a city in Illinois?). Geotagging has the ability to distinguish ambiguous place names and find the latitude and longitude of the exact location by automatically detecting clues in a given text (for example, the text mentions place names such as Alexandria and Israel). Once the precise location is identified, then it can be plotted on a map visualization tool. For more information on geotagging, see here.

Entity Extraction is an Indispensable Tool for Monitoring Geopolitical Events

In this interconnected world, organizations need highly sophisticated and reliable AI tools to navigate ongoing and emerging conflicts and threats. AI-based Entity Extraction provides an indispensable capability for companies and other organizations to monitor in real time events world-wide that might affect them in a significant way.